I’m seeking to interpret the results of a logistic regression analysis with Plink.
I'm analyzing data from a genetic case-control association study (of unrelated individuals), to try to determine association between SNPs and a disease, using Plink. I performed a standard allelic chi-square test (--assoc), and found several SNPs in one locus with significant p values. Next, I wanted to find out if these associations are independent from each other, so I used logistic regression to condition on the SNP with the lowest p value, let’s call it rs1 (i.e. the command I used is --logistic --condition rs1). After conditioning, all the rest of the SNPs that were previously significant became non-significant. Am I correct to deduce that the association found in this locus is only driven by rs1, and the significant associations in all the other SNPs are only due to their linkage disequilibrium to rs1? If I understood well all what I read about logistic regression, I think this would be the correct interpretation, but I find it a bit strange, because my SNPs are in very weak LD between them (r2<0.1)…
What strikes me more is that, in some cases, the p values corresponding to rs1 under the column TEST of the output I get also become non significant in this analysis… If rs1 is the main SNP driving association in this locus, shouldn’t it remain significant when taking the effects of the other SNPs into account?
I hope I explained my questions clearly… Any input from people experienced with genetic analysis would more than welcome. Thank you very much!